Robust Detection of Infrared Dim and Small Targets under Complex Backgrounds

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This paper is based on the inspection scenario of photovoltaic power stations and has constructed an optimized technical system for precise detection. It focuses on breaking through the key technical difficulties of precise target detection for unmanned aerial vehicles in complex terrain environments, providing a systematic solution to improve the efficiency and accuracy of photovoltaic power station inspections. For detection problems such as the weak texture features of photovoltaic modules, a target detection algorithm based on the improved YOLOv8 network is proposed. Firstly, the Swin Transformer module is introduced to construct hierarchical feature representation and sliding window mechanism, effectively capturing multi-scale information in the image. Secondly, a progressive feature pyramid network is designed, gradually transforming and integrating information from low-level features to high-level features, solving the difficulties in detecting small targets of photovoltaic modules and the problem of blurred feature loss. Finally, through simulation and experiments, the effectiveness of the proposed path planning algorithm and target detection algorithm is verified.